Headless proof of concept (with session state tracking):
Conclusion: Lot of potential but will be our “V2”. Still a lot of work to figure out the node equivalent of this capability. We’re not changing our current roadmap to incorporate this anytime soon.
Reworked node responses to separate next output ID from node output.
A lot more flexible, nodes can pass content through while selecting output edge separately.
Reworked actor/session data memory to dedicated model